I built llm-use, an open-source system for routing, caching, and A/B testing between large language models like GPT-4, Claude, Llama, and models provided via Ollama. It intelligently routes prompts across these models based on complexity and usage, helping optimize cost and performance in production.

Would love your feedback!